CICAAR: Convolutive ICA with an Auto-Regressive Inverse Model

Mads Dyrholm, Lars Kai Hansen

AbstractWe invoke an auto-regressive IIR inverse model for convolutive ICA and derive expressions for the likelihood and its
gradient. We argue that optimization will give a stable inverse. When there are more sensors than sources the mixing
model parameters are estimated in a second step by least squares estimation. We demonstrate the method on synthetic
data and finally separate speech and music in a real room recording.
KeywordsConvolutive ICA AR speech deconvolution maximum likelihood
TypeConference paper [With referee]
ConferenceIndependent Component Analysis and Blind Signal Separation
EditorsCarlos G. Puntonet and Alberto Prieto
Year2004    Month September    Vol. 3195    pp. 594-601
SeriesLecture Notes in Computer Science
ISBN / ISSN3-540-23056-4
Electronic version(s)[pdf]
BibTeX data [bibtex]
IMM Group(s)Intelligent Signal Processing

Back  ::  IMM Publications